Ph.D. Research Proposal of Jayant Apte on Algorithms for Computing Limits on Information Flow and Storage in Networks

 

Advisor

Dr. John Walsh

 

Abstract

Network coding is a communication paradigm that is used to transfer and store information across networks. In network coding, the nodes in the network are allowed to code i.e. combine incoming data using linear/non-linear functions as opposed to simply forwarding it to other nodes (routing). Network coding is known to outperform routing even in the simplest of networks. A code for a distributed storage system with certain reliability and repair requirements can also be thought of as a network coding scheme. While there exists an implicit characterization of rate region (set of all rate tuples achievable using some network coding scheme) in terms of a set that is unknown in general, explicit expressions of network coding rate regions remain unknown in general, forming a long standing open problem.

 

The proposed tasks include a rational arithmetic C implementation of symCHM algorithm, a new algorithm for isomorph free exhaustive generation of non-linear network codes and an investigation of the higher symmetries of the entropy function.